| Thermal barrier coatings (TBCs) are protective multi-layered metal-ceramic coatings used in hot sections of jet engines and gas turbines. The TBCs are composed of a superalloy substrate, an intermediate metallic bond coat (BC) and a ceramic topcoat. The TBCs are beset by reliability problems arising from delamination of the ceramic topcoat due to various instabilities in the system. The present work examines one such instability of "rumpling", or progressive roughening of the BC surface in the BC-superalloy systems upon high temperature exposure. A combined experimental and analytical approach is taken to study the rumpling phenomenon. Thermal cycling and isothermal experiments are carried out in air and in vacuum to identify the driving force and the kinetics governing rumpling. The experiments show that a nominally flat BC surface rumples to a wavelength of about 60--100 mum, and an amplitude of about 4--8 mum. The rumpling is seen to be relatively insensitive to the initial BC surface morphology. Significant initial flaws are not necessary for rumpling to occur. Further, rumpling occurs even in absence of thermal cycling. To explain BC rumpling, we develop a linear stability model for surface evolution of BCs under a remote stress. The driving force for this process is the in-plane stress in the BC due to its thermal mismatch with the substrate as indicated by the experimental results. The BC volume and BC surface diffusion governs the deformation kinetics. A governing equation is derived that gives the amplitude evolution of BC surface perturbations as a function of time. The analysis establishes a range of wavelengths for which the perturbation amplitude increases at a significantly higher rate as compared with other wavelengths. At the dominant instability wavelength, under low-stress and high-temperature conditions, the model shows that the roughening is caused only by volume diffusion, while smoothing is caused only by surface diffusion. The results from this thermodynamic model agree with the experimental observations quite well. Particular BC material properties and testing conditions are identified that control the BC rumpling and hence an important TBC failure mode. Guidelines to improve TBC performance are presented. |
